ARTICLE

The eyeless Mouse Mutation (ey1) Removes an AlternativeStart Codon from the Rx/rax Homeobox GenePriscilla Tucker,1 Lois Laemle,2 Amanda Munson,3 Shami Kanekar,4 Edward R. Oliver,3Nadean Brown,5 Hans Schlecht,2 Monica Vetter,4 and Tom Glaser3*1Museum of Zoology and Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, Michigan2Departments of Anatomy, Cell Biology and Molecular Medicine and Ophthalmology, University of Medicine andDentistry of New Jersey, Newark, New Jersey3Departments of Internal Medicine and Human Genetics, University of Michigan, Ann Arbor, Michigan4Department of Neurobiology and Anatomy, University of Utah, Salt Lake City, Utah5Children’s Memorial Institute for Education and Research, Department of Pediatrics, Northwestern University,Chicago, Illinois

Received 1 August 2001; Accepted 2 August 2001Published online 00 Month 2001; DOI 10.1002/gene.10003

Summary: The eyeless inbred mouse strain ZRDCT haslong served as a spontaneous model for human anoph-thalmia and the evolutionary reduction of eyes that hasoccurred in some naturally blind mammals. ZRDCT micehave orbits but lack eyes and optic tracts and havehypothalamic abnormalities. Segregation data suggestthat a small number of interacting genes are responsi-ble, including at least one major recessive locus, ey1.Although predicted since the 1940s, these loci werenever identified. We mapped ey1 to chromosome 18using an F2 genome scan and there found a Met103Leumutation in Rx/rax, a homeobox gene that is expressedin the anterior headfold, developing retina, pineal, andhypothalamus and is translated via a leaky scanningmechanism. The mutation affects a conserved AUGcodon that functions as an alternative translation initia-tion site and consequently reduces the abundance of Rxprotein. In contrast to a targeted Rx null allele, whichcauses anophthalmia, central nervous system defects,and neonatal death, the hypomorphic M10L allele is fullyviable. genesis 31:43–53, 2001. © 2001 Wiley-Liss, Inc.

Key words: Rx; rax; homeobox; eye development; startcodon; mouse; mutation; protein translation; leaky scanning

INTRODUCTION

The phenotype of ZRDCT mice (Fig. 1a) is similar tohuman clinical anophthalmia (Duke-Elder, 1964; Kohn etal., 1988) and the rudimentary development of eyes insome fossorial (subterranean) mammals (Cooper et al.,1993). More than 90% of mice in this inbred strain haveno eyes or optic tracts, although the eyelids, bony orbits,palpebral conjunctivae, Harderian glands, and periocularsoft tissues are intact (Chase and Chase, 1941; Silver andHughes, 1974). Higher visual centers in the brain, includ-ing the lateral geniculate nuclei and visual cortex, formnormal projections and exhibit normal topographic con-

nectivity but are attenuated (Chase, 1945; Godement etal., 1979; Katz et al., 1981; Olavarria and van Sluyters,1984; Rhoades et al., 1985). The hypothalamus has vari-ably distorted anatomy (Laemle and Rusa, 1992; Silver,1977) and there are intrinsic circadian rhythm distur-bances (Faradji-Prevautel et al., 1990; Laemle and Otten-weller, 1998). The remaining 10% of ZRDCT mice haveone or two small eyes with colobomatous defects and nooptic nerve (Silver et al., 1984). The phenotype is firstapparent during lens induction at day E10. The opticvesicles emerge normally from the anterior neural tubebut are reduced in size and contact poorly with thesurface ectoderm (Chase and Chase, 1941; Harch et al.,1978; Silver and Hughes, 1974; Webster et al., 1984). Insome embryos, there is a small displaced lens placodeand limited crystallin synthesis (Konyukhov et al., 1978;Zwaan and Silver, 1983). The eyeless trait is one of thefirst developmental defects and first examples of multi-factorial inheritance described in laboratory mice.

In crosses between ZRDCT and most other laboratorystrains, the eyeless phenotype behaves as a polygenictrait. F1 mice are normal, but 10–13% of F2 intercrossand 20–25% of N2 backcross progeny are affected (Beck,1963; Chase, 1942). On the basis of these experiments,Chase (1944) postulated a major recessive eyeless deter-minant (ey1) and a modifier (ey2) with partial domi-nance. In this study, we test this model, identify the ey1mutation, and characterize its molecular properties indetail.

* Correspondence to: Tom Glaser, 4510 MSRB I, Box 0650, 1150 WestMedical Center Drive, Ann Arbor, MI 48109-0650.

E-mail: tglaser@umich.edu

© 2001 Wiley-Liss, Inc. genesis 31:43–53 (2001)

RESULTS

To map ey1, we performed a pooled genome scan (Tay-lor et al., 1994) using F2 progeny derived from ZRDCTand CAST/Ei, an inbred strain of the relatively divergentsubspecies Mus musculus castaneus. Among 2,316 F2offspring, we observed 2,109 normal (91.1%), 153 inter-mediate (6.6%), and 54 bilaterally eyeless (2.3%) mice.The proportion of affected F2 progeny is thus between1/8 and 1/16. These ratios are similar to segregation datareported for crosses between ZRDCT and conventionallaboratory mice (Beck, 1963; Chase, 1942; Chase, 1944).By screening two pools of bilaterally anophthalmic F2mice with multiple SSLP (simple sequence length poly-morphism) markers, we detected linkage between ey1and chromosome 18. We then singly tested all 54 eyelessF2 mice using additional markers and a fragment lengthpolymorphism in the second intron of Rx/rax, a retinalhomeobox gene previously mapped to this region (Fu-rukawa et al., 1997) (Fig. 1b). No recombination wasobserved between Rx and ey1, suggesting that they arelocated within 2.8 cM (95% CI for n 5 108 meioses, witha LOD score in favor of linkage .30).

The Rx gene is an attractive candidate for ey1 forseveral reasons. First, it encodes a 342 amino acid tran-scription factor with four evolutionarily conserved mo-tifs: a paired-class homeodomain, a C-terminal OAR(paired tail) domain (Furukawa et al., 1997; Mathers etal., 1997), an octapeptide motif near the N-terminus, anda segment of high sequence conservation just beyondthe homeobox termed the Rx domain (Fig. 2a). MultipleRx homologues are known in the zebrafish, medaka,chicken, and frog, and these can be divided into twoparalogous groups (Casarosa et al., 1997; Chuang et al.,

1999; Deschet et al., 1999; Ohuchi et al., 1999). Home-odomain genes of this type are well known to regulatemorphogenesis. Second, Rx is one of the earliest retinalpatterning genes to be expressed, preceding Pax6 andChx10. It first appears at day E7.5 in regions of theanterior neural plate and headfold that are fated to be-come retina and is expressed most intensely in the opticvesicle and cup between days E9 and E11 (Fig. 1c). Rx isrelatively specific to the developing eye but also appearsin the ventral forebrain, presumptive hypothalamus, andposterior pituitary (Furukawa et al., 1997; Mathers et al.,1997). This expression pattern has been generally wellconserved during vertebrate evolution. A Drosophilahomologue is expressed in the developing brain andclypeolabrum but does not have a major role in eyemorphogenesis (Eggert et al., 1998). Third, Rx is re-quired for optic vesicle formation. Mice homozygous foran Rx null allele are anophthalmic but also die at birthwith major central nervous system and craniofacial de-fects (Mathers et al., 1997).

To evaluate Rx as a candidate for ey1, we determinedits genomic structure and DNA sequence in ZRDCT,CAST/Ei, and two conventional laboratory strains (Fig.2b). The mouse Rx gene spans 5.6 kb and is divided intothree exons. We discovered a single nucleotide changein the coding region that is unique to ZRDCT mice. Thechange (ATG3TTG) causes an M10L amino acid substi-tution and destroys an NcoI restriction site (Fig. 2b, c).To confirm the relationship between Rx(M10L) and ey1,we surveyed a variety of laboratory strains, wild-caughtmice, and other Mus species by PCR and NcoI digestion(Fig. 2d). The M10L mutation was found in ZRDCT micefrom different sources but was absent from all othermice tested, including C57BL strains which have a 4-10%

FIG. 1. Mapping ey1 in relation toRx/rax. (a) Phenotype of theZRDCT mice. (b) Segregationdata from 33 eyeless (ZRDCT 3CAST/Ei) F2 progeny localize ey1on chromosome 18 near Rx/rax.Recombinant and nonrecombi-nant chromosomes are repre-sented vertically, along with thetotal number observed in eachclass. A genetic map of MMU18derived from these data is shownon the right. Distances (cM 6 SE)are similar to those reported pre-viously (Dietrich et al., 1994). Anadditional 21 eyeless F2 mice weretyped at Rx/rax and D18Mit184. Norecombination was observed be-tween these two markers and ey1.f, ZRDCT; M, CAST/Ei genotypes.(c) Whole-mount in situ hybridiza-tion showing Rx/rax mRNA in theoptic vesicle of E9.5 and optic cupof E10.5 embryos. Expression inthe optic cup is confined to the pre-sumptive neural retina and doesnot extend into the choroid fissure.

44 TUCKER ET AL.

background frequency of anophthalmia (Chase, 1942;Pierro and Spiggle, 1967; Robinson et al., 1993; Smith etal., 1994) and may share a common ancestry (Beck et al.,2000; Chase, 1944) with the extinct B strain that is theprogenitor of ZRDCT. Moreover, a methionine codon ispresent at position 10 in Rx genes of all other mammalsexamined, including humans, rats, and six additionalphylogenetically diverse rodent species, and at theequivalent position in Rx genes of fish, amphibians, andbirds (Fig. 3a, left, and data not shown). This methioninecodon thus exhibits a level of evolutionary sequenceconservation exceeding that found among vertebrate Rxhomeodomains, which are $96% identical. It is thusimprobable that the L10 allele is segregating withinmouse populations as a neutral polymorphism. We con-clude that Rx(M10L) is most likely the ey1 determinant.

In principle, the M10L mutation could affect intrinsicproperties of the Rx protein or the efficiency of transla-tion initiation. Although this methionine codon appearsto have been selectively retained among vertebrates forat least 400 million years, the mutation itself is a conser-vative amino acid substitution (an exchange betweenhydrophobic residues) and does not occur within a pre-viously defined functional domain. The identity of the Rxstart codon is also unclear, since the sequence surround-ing the first in-frame methionine codon (M1) conformspoorly to the consensus for translation initiation com-pared with M10. In particular, the M1 site lacks a G atposition 14 and a purine at position 23, which areconsidered critical for ribosome recognition via a cap-dependent scanning mechanism (Kozak, 1996). Thesethree sequence characteristics are common to all verte-brate Rx genes (Fig. 3a, right). The choice of initiationsite can also be significantly influenced by mRNA sec-ondary structure (Kozak, 1990). To determine the trueinitiation site and the effect of the M10L mutation invivo, we compared the size of Rx-myc fusion proteins

FIG. 2. The Rx(M10L) mutation. (a) Struc-ture of the mouse Rx protein showing fourconserved motifs. (b) Genomic map of Rxshowing three exons, the 3-bp fragmentlength polymorphism (FLP) used for link-age mapping, the PCR primers used formutation screening (blue arrows), and theM10L mutation. Homeobox segments(yellow) are indicated within the codingregion (gray). The relevant ZRDCT nucle-otide sequence is shown below the wildtype, along with the NcoI site (red text).Complete genomic sequence data areavailable from Genbank. (c) Southernanalysis of Rx confirming the loss of anNcoI restriction site in ZRDCT mice. C,C3H/HeJ; B, C57BL/6J; Z, ZRDCT. (d)Survey of selected mouse strains for theRx(M10L) mutation. PCR products fromexon1 were digested with NcoI and elec-trophoresed through a 3% agarose gel.Similar results were obtained with DNAsamples from other strains and wild-caught mice (see Methods).

45RX/RAX MUTATION IN THE EYELESS (EY1) MOUSE

translated in cells transfected with mutant or wild-typeplasmid constructs (Fig. 3b). The fusion mRNAs contain59 UTR sequences and the first 66 potential codons ofthe mouse Rx gene. Proteins initiating at codons 1 or 10are predicted to be 18.8 or 17.9 kDa, respectively. Wealso tested control constructs containing an M1L substi-tution or an optimal initiation sequence at M1. When thewild-type fusion construct was tested, both methioninecodons were utilized as translation start sites (lane 6),

consistent with a leaky scanning model (Kozak, 1995).The relative abundance of these two proteins presum-ably reflects the balance between competing factors thatfavor initiation at M1 (proximity to the 59 m7G cap)versus M10 (context superiority for engaging the 40Sribosomal subunit). The M10L mutation prevented initi-ation at codon 10 and significantly lowered the effi-ciency of Rx translation overall (compare lanes 5 and 6).The decrease in translation efficiency was demonstrated

FIG. 3. The Rx(M10L) mutation affects translation initiation. (a) Left: N-terminal alignment of vertebrate Rx proteins. Sequences wereobtained from GenBank and are numbered in relation to mouse Rx/rax. Methionines (red) and residues that differ from mouse Rx/rax (blue)are indicated. The methionine at position 9, 10, or 11 has been conserved during evolution. Right: Comparison of M1 and M9–11translational start sites in vertebrate Rx mRNAs. The optimal sequence for translation initiation (Kozak, 1996) is shown below eachalignment, and the nucleotide matches are indicated in red text. In each case, the downstream site is a significantly more favorable contextthan M1. (b) Western analysis of Rx-myc fusion polypeptides translated in vivo (transfected 293T cells). Left: Western blot showingMyc-tagged proteins initiating at M1, M10, or M73 (within the pCS2MT vector). Lanes 1 and 2 contain mixtures of cell extracts from lanes315 and lanes 415, respectively. The a-tubulin blot below is a loading control. Right: Magnified view of Myc Western blot (lanes 4–6). Thediagram below shows the framework pCS2MT expression construct and predicted protein products. Wild-type Rx mRNA starts translationat both M1 and M10 codons. The M1. mRNA starts translation exclusively at M1, whereas the M1L mRNA starts translation exclusivelyat M10. The M10L mutation eliminates the shorter product and decreases overall translation efficiency. MT, myc tag; ., optimal start site.

46 TUCKER ET AL.

most clearly by the appearance of an 11.1 kDa immuno-reactive protein; this reflects ribosome scanning pastboth codons 1 and 10 and initiation at the first in-frameAUG codon of the pCS2MT vector (compare lanes 5 and7). In contrast, the reciprocal M1L mutation preventedinitiation at codon 1 in vivo but did not significantlylower overall translation (lane 4).

To assess the effect of the M10L mutation quantita-tively in the context of the full-length Rx transcript, wecompared Rx protein levels in fetal retinal cells thatwere transfected with either a mutant or wild-type ex-pression plasmid (Fig. 4a). These plasmids express theentire Rx mRNA, including the open reading frame andUTRs, but differ by a single nucleotide. We observed asignificant reduction in the abundance of Rx protein(compare lanes 3–6), which can best be explained by adecrease in translation efficiency. The mutation does notdestabilize Rx mRNA, since wild-type and M10L tran-scripts were equally abundant in this experiment (data

not shown) and in heterozygous (C3H/HeJ 3 ZRDCT) F1embryos (Fig. 4b).

Finally, as a further functional test, we compared theactivity of Rx isoforms in a Xenopus RNA microinjectionassay (Andreazzoli et al., 1999; Mathers et al., 1997). Wefirst tested an M9L substitution in frog Xrx1 cDNA that isequivalent to the ZRDCT mouse Rx mutation. Wheninjected into a D.1 blastomere of an 8-cell frog embryo,M9L and wild-type Xrx1 RNAs were both capable ofinducing ectopic pigmented epithelium, although nei-ther caused significant expansion of neural retina (datanot shown). We then compared the activity of mouse Rxproteins initiating at M1 or M10 and M1-initiated pro-teins with methionine or leucine at codon 10. We ob-served a wide range of abnormal phenotypes on theinjected sides, including small or absent eyes, cyclopia,retinal disruption, and ectopic pigmented epitheliumalong the optic stalk (data not shown). These resultsconfirm that eye morphogenesis is sensitive to Rx dos-

FIG. 4. The M10L mutation reduces the abundance of Rxprotein, but not Rx mRNA. (a) Western analysis of HER10retinal cells expressing full-length wild-type or M10L mutantRx mRNAs. Top: The level of Rx polypeptide (36 kDa, arrow)is decreased approximately four-fold in duplicate culturestransfected with pCS2(RxM10L) compared with wild-typepCS2(Rx) plasmid. The M1- and M10-initiated Rx isoformsare not resolved in this gel. Bottom: Endogenous a-tubulin(57 kDa) shows equivalent protein loading. (b) M10L andwild-type Rx transcripts are amplified equally from E14.5(C3H/HeJ 3 ZRDCT) F1 embryos. An RT-PCR spanningexons 1–3 was performed on total head RNA from fourlittermates (lanes 1–4). The wild-type product is cleaved byNcoI (554 bp) but the mutant product is resistant (824 bp).These alleles are equally represented in the heterozygotes,suggesting that the mRNAs are similarly expressed and sta-ble. No product was amplified in the absence of reversetranscriptase.

47RX/RAX MUTATION IN THE EYELESS (EY1) MOUSE

age. (Fig. 5) (Table 1). However, there was no significantqualitative difference between the mutant and wild-typeisoforms or between proteins initiating at M1 or M10 intheir effect on frog eye development. The ey1 mutationthus does not grossly alter the intrinsic biological prop-erties of Rx proteins.

DISCUSSION

Taken together, our findings show that ribosomes fre-quently bypass the suboptimal M1 site and initiate at thedownstream M10 codon, producing two Rx proteinsthat differ by nine amino acids from one mRNA. Al-though rare, translational control by alternative startcodon usage from a single transcript can provide a fur-ther opportunity to regulate protein function (Corneliset al., 2000; Hann et al., 1988; Liu et al., 1997; Markussenet al., 1995; Sedman and Mertz, 1988). While this re-mains a possibility for Rx, and the M10L mutation doesdelete the shorter isoform, we believe the primary effectof ey1 is a quantitative reduction in the synthesis of Rxpolypeptide. This effect was likely to be significantlygreater than two-fold a priori, since mice heterozygousfor a targeted Rx null allele are phenotypically normal(Mathers et al., 1997). Indeed, our data suggest that theM10L mutation decreases Rx translation by at least four-fold (Figs. 3b, 4a). This result, together with (1) themeiotic linkage between Rx and ey1; (2) the uniqueM10L sequence alteration in the ZRDCT strain; (3) thedeep evolutionary conservation of M10; (4) the estab-lished role of Rx and homeodomain proteins generally ineye development; and (5) the specific effect of M10L onalternative initiation, strongly suggests that Rx(M10L)can be equated with ey1. A potentially similar but lessprofound reduction in the efficiency of translation initi-ation was reported for a mutation in codon 2 of thehuman androgen receptor gene that causes a hypomor-phic phenotype (Choong et al., 1996). In contrast, mis-sense mutations affecting solitary start codons generallycause a complete loss of function (Cheadle et al., 1994;Tsujino et al., 1994; Waye et al., 1997). To our knowl-

edge, Rx(M10L) is the first example of a constitutionalmutation affecting an alternative start codon.

Our findings explain a number of previous observa-tions regarding Rx and this strain. First, the eye andhypothalamic phenotypes of ZRDCT mice closely matchthe Rx expression pattern. Second, the M10L substitu-tion is compatible with the behavior of ey1 as a hypo-morphic allele. This is manifest during embryogenesis bya graded reduction in the size of the optic vesicle and inthe adult phenotype. Third, explant data show an intrin-sic growth defect in the neural retina of ZRDCT embryosbut not in lens primordia (Koniukhov and Iakovlev,1986). Within the mouse eye, Rx expression is limited tothe optic cup and neural retina and is relatively indepen-dent of Pax6 (Bernier et al., 2000; Zhang et al., 2000; NBand TG, unpublished observations). Accordingly, mi-crophthalmia or anophthalmia can result from mutationsin patterning genes that are expressed in the optic cup,lens, or both (Breitman et al., 1987; Burmeister et al.,1996; Ferda Percin et al., 2000; Hill et al., 1991; Hodgkin-son et al., 1993).

Approximately one-half to two-thirds of the F2 micehomozygous for the Rx(M10L) mutation have normallysized eyes. This conclusion follows from the phenotypicsegregation data, which show fewer affected mice thanthe 25% expected for a single recessive mendelian fac-tor; the absence of excess perinatal lethality among theF2 progeny and parental ZRDCT mice; and the finding ofM10L homozygotes in the normal F2 progeny class (datanot shown). The Rx(M10L) mutation is thus necessarybut not sufficient to produce anophthalmia in theZRDCT strain. This decreased penetrance and the phe-notypic variability among affected F2 mice are reminis-cent of anophthalmia and microphthalmia in some hu-man pedigrees (Bateman, 1984). Our results suggest thatRX mutations may underlie some of these human disor-ders (Voronina and Mathers, 2000). The inheritance pat-tern also suggests that unlinked modifier gene(s), arelikely to interact with Rx in a developmental pathway forthe eye or otherwise control the threshold for pheno-typic expression of ey1. Although a single modifier (ey2)

FIG. 5. Phenotypes resulting from overexpression of mouse Rx in Xenopus (shown at stages 39–41). RNA was injected into one dorsalblastomere of each 8-cell embryo. (A) Abnormal pigmentation; (B) missing eye; (C) eyes joined at midline

48 TUCKER ET AL.

was originally proposed by Chase (1944), the actualnumber of modifiers cannot be determined from segre-gation data alone (Beck, 1963; Wright, 1968), sincetheir individual dominance characteristics and poten-tial epistatic relationships are unknown. This number,however, is likely to be small. The discovery of ey1provides a basis to identify these genes. Comparablemodifier loci have been defined for mutations affect-ing limb morphogenesis (Johnson et al., 1995; Sidowet al., 1997) and other eye diseases (Bone-Larson et al.,2000; Chang et al., 1999; Kajiwara et al., 1994).

Finally, as a synthetic trait involving a partial loss-of-function mutation, the ZRDCT phenotype is suggestiveof the quantitative, polygenic interactions that underliethe appearance and loss of morphological characteristicsduring evolution (Grenier and Carroll, 2000; Wright,1968). Extreme eye reduction is typical of troglobitic(cave-dwelling) and fossorial species, which are foundwithin every class of vertebrates except birds. Adaptiveloss of function (Olson, 1999) involving quantitativechanges in the activity of genes such as Rx may thuscontribute to ocular regression in naturally blind species(Cooper et al., 1993; Wilkens, 1971; Yamamoto andJeffery, 2000).

METHODS

MiceThe inbred ZRDCT strain (aa bb dd pp) was derived

(Beck, 1963) from the original B strain (bb dd) of Her-man Chase, who received founders directly from C.C.Little in 1938 (Chase, 1942). The eyeless phenotype wasfirst noted at the Jackson Laboratory within a colony ofstrain R mice, among progeny of a wide cross, and waspositively selected by Little during the initial breeding(Chase, 1949; Snell, 1941). To generate intersubspecific

F2 mice, we crossed eyeless ZRDCT males and femalesto inbred Mus musculus castaneus mice (CAST/Ei, ob-tained from the Jackson Laboratory) and intercrossed theresulting F1 progeny. Phenotypes were scored twoweeks after birth when the eyelids had opened. Eyeswere classified according to Chase (1942) as normal,small, or absent, and individual F2 pups as normal, eye-less (bilateral anophthalmia), or intermediate (microph-thalmia or unilateral anophthalmia). In cases where thelids remained closed, the conjunctival sacs were ex-plored and generally found to be anophthalmic. For RNAanalysis, we crossed ZRDCT males to C3H/HeJ femalesand recovered F1 embryos on day E14.5 of gestation.

PCR Genotyping

A genome scan was performed using 58 SSLP mark-ers distributed across the autosomes (Dietrich et al.,1994). Genomic DNA from 12 severely affected F2mice was divided into two equal pools (Taylor et al.,1994). Markers were amplified for 40 cycles usingstandard PCR conditions and primer pairs (ResearchGenetics). Short, secondary PCRs (10 cycles) werethen performed with one 32P-end-labeled primer andthe products separated by electrophoresis through 6%polyacrylamide sequencing gels. F2 pools were com-pared with ZRDCT, CAST/Ei, and F1 genomic DNAcontrols. Once linkage was detected, by an enrich-ment of ZRDCT alleles in the pooled DNA, a larger setof 54 eyeless F2 mice was typed individually usingadditional markers on chromosome 18.

Rx Gene Analysis

Overlapping segments of mouse Rx were amplifiedfrom ZRDCT, C3H/HeJ, C57BL/6J, and CAST/Ei genomicDNA by high-fidelity long-range PCR (EXPAND, Roche)using primers designed from the cDNA sequence (Fu-rukawa et al., 1997; Mathers et al., 1997). Betaine (20%v/v MasterAmp, Epicentre) was included in some reac-tions to improve yield of GC-rich products. Terminalfragments were recovered from BamHI-digestedgenomic DNA by inverse PCR (Ochman et al., 1990).The amplified products were sequenced directly andused to assemble two Rx contigs. These are separated by2.1 kb and correspond to exons 1–2 and 3 (accessionnos. AF319462 and AF319461). To localize Rx directly inthe F2 cross, we typed a 3-bp insertion in the secondintron, using PCR primers 59-CCTGTGGGTCAGAGAG-GATAGCGAC-39 and 59-CTGGCGCCTCCACTTAGCCCG-TCGG-39 which span 265 bp. Southern analysis of mousegenomic DNA was performed using a 785-bp 59 RxcDNA probe (Mathers et al., 1997) and high stringencywash conditions (0.1 3 SSC, 65°C). To demonstrate theM10L Rx mutation, a 361-bp fragment of exon 1 wasamplified using primers 59-CTAAACTTGCAGCTCCAG-CAGCGGG-39 and 59-GCCCAGGATGGCTTCGATGCT-GTG-39 and digested with NcoI. The wild-type PCR prod-uct is cleaved into 269- and 92-bp fragments, but theM10L product is resistant to cleavage. We used this assayto survey normal inbred laboratory mice (A/J, AU/SsJ,

Table 1Phenotypes Observed Following Overexpression of Mouse Rx

in Xenopus Embryos*

Phenotype

Rx(wild type) Rx (M10L)

No. % No. %

Pigmentation abnormalities 114 62 87 39Small or missing eyes 30 16 36 16Eyes joined along midline 6 3 9 4Other abnormal phenotypesa 3 2 11 5Normal 32 17 81 36Total 185 224

*RNA (10 pg) was transcribed in vitro from wild-type or M10LpCS2(Rx) plasmids, injected into one dorsal blastomere of an 8-cellembryo, and evaluated at stage 39–41. Although there was noqualitative difference between mutant and wild type, the M10L RNAwas generally less potent, consistent with decreased translationefficiency. Similar results were observed with wild-type and M9LXrx1 RNAs.

aOther phenotypes we observed include: no eyes, with randompigmentation along the midline of the anterior head (most common);one eye displaced anteriorly at midline; and no eyes, with no pig-mentation.

49RX/RAX MUTATION IN THE EYELESS (EY1) MOUSE

BALB/cJ, CBA/J, C3H/HeJ, C57BL/6J, C57BL/10J, C57BR/cdJ, C57L/J, C58/J, DBA/2J, LT/ChRe, SJL/J, SWR/J, 129/SvJ); Mus musculus castaneus (CAST/Ei), Mus spretus(SPE/Ei), Mus musculus molossinus (MOLF/Ei), and aPeruvian strain (PERA), each obtained as DNA from theJackson Laboratory; wild-caught Mus musculus and Musdomesticus from the hybrid zone of Northern Europe(kindly provided by Richard Sage); and eyeless ZRDCTmice from colonies in Rhode Island (Harch et al., 1978)and Moscow (Konyukhov et al., 1978), which have beenmaintained separately for more than 20 years (kindlyprovided by Neil Gonsalves and Boris Konyukhov).

RNA Analysis

Whole-mount in situ hybridization was performed onwild-type CD1 mouse embryos as described (Brown etal., 1998) using a 59 Rx cDNA probe (Mathers et al.,1997). To test whether the M10L mutation affects tran-script stability, we performed a reverse transcriptase(RT) PCR assay on (C3H/HeJ 3 ZRDCT) F1 embryos.Total RNA from E14.5 heads was randomly primed andused to amplify 824-bp Rx cDNA products spanningexons 1–3 with PCR primers 59-CTAAACTTGCAGCTC-CAGCAGCGGG-39 and 59-CTGGCGCCTCCACTTAGC-CCGTCGG-39. The reverse primer was 32P-end-labeledand the PCR was for 30 cycles. Mutant and wild-typetranscripts differ by a single nucleotide. Their relativeabundance was determined by cleaving the radiolabeledPCR products with NcoI and comparing the ratio of 824-and 554-bp allelic fragments after polyacrylamide gelelectrophoresis. The two alleles amplified equally in con-trol reactions performed on F1 genomic DNA.

Rx Plasmids

Mouse Rx cDNA clones with M10L and M1L mutations(ATG3TTG) or optimal translation initiation sites(Kozak, 1996) M1. and M10. (GCCACCATGG) werecreated by oligonucleotide mutagenesis and restrictionendonuclease selection (Deng and Nickoloff, 1992). AnM9L mutation was introduced by the same method intoa frog Xrx1 cDNA clone in the pSP64RI vector (kindlyprovided by Pete Mathers). Full-length mouse Rx expres-sion constructs were created in pCS2 by ligating XbaI-BsrGI fragments from 59 and 39 cDNA clones (Mathers etal., 1997). To determine the exact site of translationinitiation, a panel of mutant and wild-type Rx-myc fusionconstructs was created in pCS2MT (Rupp et al., 1994;Turner and Weintraub, 1994). The expression cassettescontain 86 bp of 59UTR and the 66 amino-terminalcodons of mouse Rx, followed in frame by six tandemMyc epitopes, and are transcribed in vivo from the sim-ian CMV promoter. Fusion polypeptides starting atcodons M1 or M10 contain 166 or 157 amino acids,respectively. The fusion constructs also contain an opti-mal translation initiation sequence (AAAGCTATGG) atM73 within the pCS2MT vector.

Cell Culture and Protein Analysis

Translation initiation was studied in vivo using 293Thuman embryonic kidney cells. Cultures were trans-fected with pCS2(Rx-myc) or control pCS2MT plasmidsby CaPO4 co-precipitation (2.5 mg per 60-mm dish).After 48 h, whole-cell extracts were harvested in coldlysis buffer (50 mM Tris, 150 mM NaCl, 1% NP40, pH8.0) and clarified by centrifugation. To resolve fusionproteins in the size range between 17 and 19 kDa,supernatants were electrophoresed through a 14-cm18% polyacrylamide gel for 10 h at 14 V/cm. The gelbuffer was 350 mM Bis-Tris, pH 6.5 and the runningbuffer was 50 mM MOPS, 50 mM Tris, 0.1% SDS, 1 mMEDTA, pH 7.7. Myc-tagged proteins were detected inWestern blots using 9E10 monoclonal antibody (1:200),HRP-conjugated sheep anti-mouse IgG (1:20,000, Amer-sham), and enhanced chemiluminescence (ECL) re-agents (Renaissance, NEN). To confirm the interpreta-tion of migration patterns, selected cell extracts frommutant and control transfections were mixed and co-electrophoresed. Similar results were obtained forpCS2(Rx-myc) constructs in three different transfectionexperiments. To verify that an equivalent amount of cellextract was loaded in each lane and control for transfec-tion efficiency, we probed nitrocellulose filters with amonoclonal antibody (DM1A) that detects endogenousa-tubulin (1:200, Neomarkers) or co-transfected a plas-mid encoding an unrelated Myc-tagged protein.

To assess the efficiency of Rx translation overall, du-plicate 60-mm cultures of Ad12 HER10 human embry-onic retinal cells (Grabham et al., 1988) were co-trans-fected in parallel with 1.0 mg of the full-length mutant orwild-type Rx expression constructs or pCS2 vector. Plas-mid DNAs were introduced using FuGENE6 reagent(Roche). After 60 h, soluble proteins were extractedfrom cells by lysis in cold RIPA buffer (150 mM NaCl,1.0% NP40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris,pH 8.0) and electrophoresed through 10% polyacryl-amide SDS Bis-Tris gels (Novex, 20 mg per lane). Aftertransfer, Western blots were probed in series with poly-clonal rabbit anti-RX IgG (1:200) and DM1A (a-tubulin).The anti-RX IgG was purified from whole rabbit serum(Kimura et al., 2000). Primary antibodies were detectedusing HRP-protein A (1:5,000, KPL) or HRP-conjugatedsheep anti-mouse IgG (1:20,000, Amersham) and ECLreagents (NEN). The relative abundance of Rx proteinwas determined by densitometry from multiple ECL ex-posures. The Rx mRNA levels in transfected cells wereequivalent by RT-PCR.

Xenopus RNA Microinjection

Capped RNA from wild-type and M9L mutant Xrx1cDNA clones was synthesized in vitro using SP6 poly-merase (Ambion) and microinjected using standard tech-niques (Kanekar et al., 1997) into one of two D.1 blas-tomeres in 8-cell Xenopus embryos. These dorsalblastomeres normally form retinae, brain, and other an-terior structures (Bauer et al., 1994). The injected Xrx1

50 TUCKER ET AL.

transcripts contain rabbit b-globin UTR sequences butare not polyadenylated. We tested a range of Xrx1 RNAamounts from 50 to 200 pg and co-injected a b-gal tracerRNA (50 pg). Similar experiments were performed using10 to 68 pg capped mouse Rx RNA derived from wild-type, M1L, M10L, M1., M10. and M1./M10L clones inpCS2 (Rupp et al., 1994; Turner and Weintraub, 1994);these transcripts contain an SV40 polyA cassette that canbe utilized by injected Xenopus embryos, leading tosignificantly higher levels of protein expression (D.L.Turner, personal communication). Embryos werestained for b-galactosidase at stage 39–41 and evaluatedunder the dissection microscope and by histology. Tox-icity was assessed by comparing the extent of staining inembryos co-injected with Rx and b-gal RNAs to embryosinjected with b-gal RNA alone. Abnormal phenotypeswere assessed in comparison with the contralateral un-injected sides as described (Mathers et al., 1997).

GenBank Accession Numbers

Mus musculus Rx/rax genomic DNA, AF319462 andAF319461; Mus musculus Rx/rax cDNA, AF001906 andU73177; Rattus norvegicus Rx genomic DNA, AF320224and AF320225; Rattus norvegicus Rx cDNA AF135839;Homo sapiens RX cDNA, AF115392; Xenopus laevisXrx1 cDNA, AF017273 and AF001048; Xenopus laevisXrx2 cDNA, AF001049; Gallus gallus cRx2/cRax2cDNA, AF092538; Danio rerio Zrx1 cDNA, AF001907;Danio rerio Zrx2 cDNA, AF001908; Danio rerio Zrx3cDNA, AF001909; Astyanax mexicanus Rx1 cDNA,AF264703; Oryzias latipes Rx2 cDNA, AJ007939.

ACKNOWLEDGMENTS

We thank Ursula Drager, Neil Gonsalves, and BorisKonyukhov for providing ZRDCT mice; Pete Mathers forRx cDNA clones; David Turner for pCS2 vectors andadvice; Toshi Shinohara for anti-RX antisera; Phil Galli-more for HER10 cells; Richard Sage for wild-caught MusDNA samples; Chuan-Min Chen for technical assistance;Karen Rader for historical advice regarding C.C. Littleand the origins of ZRDCT mice; Ursula Drager, PeteMathers, Jerry Silver, David Ginsburg, Pam Raymond,John Moran, Kevin Patrie, Ben Margolis, and CharlesMitchell for helpful advice; and Miriam Meisler, MargitBurmeister, and Sally Camper for critically reading themanuscript. This work was supported by grants from theNSF to PKT, from the NIH and Howard Hughes MedicalInstitute to TG, from Research to Prevent Blindness; Inc.to LL, and from the NIH and Pew Charitable Truststo MV.

LITERATURE CITED

Andreazzoli M, Gestri G, Angeloni D, Menna E, Barsacchi G. 1999. Roleof Xrx1 in Xenopus eye and anterior brain development. Devel-opment 126:2451–2460.

Bateman JB. 1984. Microphthalmos. Int Ophthalmol Clin 24:87–107.Bauer DV, Huang S, Moody SA. 1994. The cleavage stage origin of

Spemann’s Organizer: Analysis of the movements of blastomere

clones before and during gastrulation in Xenopus. Development120:1179–1189.

Beck JA, Lloyd S, Hafezparast M, Lennon-Pierce M, Eppig JT, FestingMF, Fisher EM. 2000. Genealogies of mouse inbred strains. NatGenet 24:23–25.

Beck SL. 1963. The anophthalmic mutant of the mouse I. Geneticcontribution to the anophthalmic phenotype. J Hered 54:39–44.

Bernier G, Panitz F, Zhou X, Hollemann T, Gruss P, Pieler T. 2000.Expanded retina territory by midbrain transformation upon over-expression of six6 (Optx2) in xenopus embryos. Mech Dev 93:59–69.

Bone-Larson C, Basu S, Radel JD, Liang M, Perozek T, Kapousta-BruneauN, Green DG, Burmeister M, Hankin MH. 2000. Partial rescue ofthe ocular retardation phenotype by genetic modifiers. J Neuro-biol 42:232–247.

Breitman ML, Clapoff S, Rossant J, Tsui LC, Glode LM, Maxwell IH,Bernstein A. 1987. Genetic ablation: Targeted expression of atoxin gene causes microphthalmia in transgenic mice. Science238:1563–1565.

Brown NL, Kanekar S, Vetter ML, Tucker PK, Gemza DL, Glaser T.1998. Math5 encodes a murine basic helix-loop-helix transcriptionfactor expressed during early stages of retinal neurogenesis. De-velopment 125:4821–4833.

Burmeister M, Novak J, Liang MY, Basu S, Ploder L, Hawes NL, VidgenD, Hoover F, Goldman D, Kalnins VI, Roderick TH, Taylor BA,Hankin MH, McInnes RR. 1996. Ocular retardation mouse causedby Chx10 homeobox null allele: Impaired retinal progenitor pro-liferation and bipolar cell differentiation. Nat Genet 12:376–384.

Casarosa S, Andreazzoli M, Simeone A, Barsacchi G. 1997. Xrx1, a novelXenopus homeobox gene expressed during eye and pineal glanddevelopment. Mech Dev 61:187–198.

Chang B, Smith RS, Hawes NL, Anderson MG, Zabaleta A, Savinova O,Roderick TH, Heckenlively JR, Davisson MT, John SW. 1999.Interacting loci cause severe iris atrophy and glaucoma in DBA/2Jmice. Nat Genet 21:405–409.

Chase HB. 1942. Studies on an anophthalmic strain of mice. III. Resultsof crosses with other strains. Genetics 27:339–348.

Chase HB. 1944. Studies on an anophthalmic strain of mice. IV. Asecond major gene for anophthalmia. Genetics 29:264–269.

Chase HB. 1945. Studies on an anophthalmic strain of mice. V. Asso-ciated cranial nerves and brain centers. J Comp Neurol 83:121–139.

Chase HB. 1949. Mouse News Lett 1:11–12.Chase HB, Chase EB. 1941. Studies of an anophthalmic strain of mice.

I. Embryology of the eye region. J Morph 68:279–301.Cheadle JP, Belloni E, Ferrari M, Millar-Jones L, Meredith AL. 1994. A

novel mutation (M1V) in the translation initiation codon of thecystic fibrosis transmembrane conductance regulator gene, inthree CF chromosomes of Italian origin. Hum Mol Genet 3:1431–1432.

Choong CS, Quigley CA, French FS, Wilson EM. 1996. A novel missensemutation in the amino-terminal domain of the human androgenreceptor gene in a family with partial androgen insensitivity syn-drome causes reduced efficiency of protein translation. J ClinInvest 98:1423–1431.

Chuang JC, Mathers PH, Raymond PA. 1999. Expression of three Rxhomeobox genes in embryonic and adult zebrafish. Mech Dev84:195–198.

Cooper HM, Herbin M, Nevo E. 1993. Visual system of a naturallymicrophthalmic mammal: The blind mole rat, Spalax ehrenbergi.J Comp Neurol 328:313–350.

Cornelis S, Bruynooghe Y, Denecker G, Van Huffel S, Tinton S, BeyaertR. 2000. Identification and characterization of a novel cell cycle-regulated internal ribosome entry site. Mol Cell 5:597–605.

Deng WP, Nickoloff JA. 1992. Site-directed mutagenesis of virtually anyplasmid by eliminating a unique site. Anal Biochem 200:81–88.

Deschet K, Bourrat F, Ristoratore F, Chourrout D, Joly JS. 1999. Ex-pression of the medaka (Oryzias latipes) Ol-Rx3 paired-like gene intwo diencephalic derivatives, the eye and the hypothalamus.Mech Dev 83:179–182.

Dietrich WF, Miller JC, Steen RG, Merchant M, Damron D, Nahf R,Gross A, Joyce DC, Wessel M, Dredge RD, Marquis A, Stain LD,

51RX/RAX MUTATION IN THE EYELESS (EY1) MOUSE

Goodman N, Page DC, Lander ES, 1994. A genetic map of themouse with 4,006 simple sequence length polymorphisms. NatGenet 7:220–245.

Duke-Elder S. 1964. Congenital deformities of the eye. I. Anomalies inorganogenesis. St. Louis, MO: CV Mosby. p 415–428.

Eggert T, Hauck B, Hildebrandt N, Gehring WJ, Walldorf U. 1998.Isolation of a Drosophila homolog of the vertebrate homeoboxgene Rx and its possible role in brain and eye development. ProcNatl Acad Sci USA 95:2343–2348.

Faradji-Prevautel H, Cespuglio R, Jouvet M. 1990. Circadian rest-activityrhythms in the anophthalmic, monocular and binocularZRDCT/An mice. Retinal and serotoninergic (raphe) influences.Brain Res 526:207–216.

Ferda Percin E, Ploder LA, Yu JJ, Arici K, Jonathan Horsford D, Ruth-erford A, Bapat B, Cox DW, Duncan AM, Kalnins VI, Kocak-Altintas A, Sowden JC, Traboulsi E, Sarfarazi M, McInnes RR. 2000.Human microphthalmia associated with mutations in the retinalhomeobox gene CHX10. Nat Genet 25:397–401.

Furukawa T, Kozak CA, Cepko CL. 1997. rax, a novel paired-typehomeobox gene, shows expression in the anterior neural fold anddeveloping retina. Proc Natl Acad Sci USA 94:3088–3093.

Godement P, Saillour P, Imbert M. 1979. Thalamic afferents to thevisual cortex in congenitally anophthalmic mice. Neurosci Lett13:271–278.

Grabham PW, Grand RJA, Byrd PJ, Gallimore PH. 1988. Differentiationof normal and adenovirus-12 E1 transformed human embryo reti-nal cells. Exp Eye Res 47:123–133.

Grenier JK, Carroll SB. 2000. Functional evolution of the Ultrabithoraxprotein. Proc Natl Acad Sci USA 97:704–709.

Hann SR, King MW, Bentley DL, Anderson CW, Eisenman RN. 1988. Anon-AUG translational initiation in c-myc exon 1 generates anN-terminally distinct protein whose synthesis is disrupted in Bur-kitt’s lymphomas. Cell 52:185–195.

Harch C, Chase HB, Gonsalves NI. 1978. Studies on an anophthalmicstrain of mice. VI. Lens and cup interaction. Dev Biol 63:352–357.

Hill RE, Favor J, Hogan BL, Ton CC, Saunders GF, Hanson IM, ProsserJ, Jordan T, Hastie ND, van Heyningen V. 1991. Mouse small eyeresults from mutations in a paired-like homeobox-containing gene.Nature 354:522–525.

Hodgkinson CA, Moore KJ, Nakayama A, Steingrimsson E, CopelandNG, Jenkins NA, Arnheiter H. 1993. Mutations at the mousemicrophthalmia locus are associated with defects in a gene en-coding a novel basic-helix-loop-helix-zipper protein. Cell 74:395–404.

Johnson KR, Lane PW, Ward-Bailey P, Davisson MT. 1995. Mapping themouse dactylaplasia mutation, Dac, and a gene that controls itsexpression, mdac. Genomics 29:457–464.

Kajiwara K, Berson EL, Dryja TP. 1994. Digenic retinitis pigmentosadue to mutations at the unlinked peripherin/RDS and ROM1 loci.Science 264:1604–1608.

Kanekar S, Perron M, Dorsky R, Harris WA, Jan LY, Jan YN, Vetter ML.1997. Xath5 participates in a network of bHLH genes in thedeveloping Xenopus retina. Neuron 19:981–994.

Katz MJ, Lasek RJ, Kaiserman-Abramof IR. 1981. Ontophyletics of thenervous system: Eyeless mutants illustrate how ontogenetic buffermechanisms channel evolution. Proc Natl Acad Sci USA 78:397–401.

Kimura A, Singh D, Wawrousek EF, Kikuchi M, Nakamura M, ShinoharaT. 2000. Both PCE-1/RX and OTX/CRX interactions are necessaryfor photoreceptor-specific gene expression. J Biol Chem 275:1152–1160.

Kohn G, el Shawwa R, el Rayyes E. 1988. Isolated “clinical anophthal-mia” in an extensively affected Arab kindred. Clin Genet 33:321–324.

Koniukhov BV, Iakovlev MI. 1986. [Expression of mutant eyeless genesin the mouse embryo retina]. Biull Eksp Biol Med 101:738–740.

Konyukhov BV, Malinina NA, Platonov ES, Yakovlev MI. 1978. Immu-nohistochemical study of crystallin synthesis during morphogen-esis of the crystalline lens in mice. Biol Bull Acad Sci USSR5:397–405.

Kozak M. 1990. Downstream secondary structure facilitates recogni-

tion of initiator codons by eukaryotic ribosomes. Proc Natl AcadSci USA 87:8301–8305.

Kozak M. 1995. Adherence to the first-AUG rule when a second AUGcodon follows closely upon the first. Proc Natl Acad Sci USA92:7134.

Kozak M. 1996. Interpreting cDNA sequences: Some insights fromstudies on translation. Mamm Genome 7:563–574.

Laemle LK, Ottenweller JE. 1998. Daily patterns of running wheelactivity in male anophthalmic mice. Physiol Behav 64:165–171.

Laemle LK, Rusa R. 1992. VIP-like immunoreactivity in the suprachias-matic nuclei of a mutant anophthalmic mouse. Brain Res 589:124–128.

Liu Y, Garceau NY, Loros JJ, Dunlap JC. 1997. Thermally regulatedtranslational control of FRQ mediates aspects of temperature re-sponses in the neurospora circadian clock. Cell 89:477–486.

Markussen FH, Michon AM, Breitwieser W, Ephrussi A. 1995. Transla-tional control of oskar generates short OSK, the isoform thatinduces pole plasma assembly. Development 121:3723–3732.

Mathers PH, Grinberg A, Mahon KA, Jamrich M. 1997. The Rx ho-meobox gene is essential for vertebrate eye development. Nature387:603–607.

Ochman H, Medhora MM, Garza D, Hartl DL. 1990. Amplification offlanking sequences by inverse PCR. In: Innis M, Gelfand D, SninskyJ, White T, editors. San Diego: Academic Press. p 219–227.

Ohuchi H, Tomonari S, Itoh H, Mikawa T, Noji S. 1999. Identificationof chick rax/rx genes with overlapping patterns of expressionduring early eye and brain development. Mech Dev 85:193–195.

Olavarria J, van Sluyters RC. 1984. Callosal connections of the posteriorneocortex in normal-eyed, congenitally anophthalmic, and neona-tally enucleated mice. J Comp Neurol 230:249–268.

Olson MV. 1999. When less is more: Gene loss as an engine ofevolutionary change. Am J Hum Genet 64:18–23.

Pierro LJ, Spiggle J. 1967. Congenital eye defects in the mouse. I.Corneal opacity in C57black mice. J Exp Zool 166:25–33.

Rhoades RW, Mooney RD, Fish SE. 1985. Subcortical projections ofarea 17 in the anophthalmic mouse. Brain Res 349:171–181.

Robinson ML, Holmgren A, Dewey MJ. 1993. Genetic control of ocularmorphogenesis: Defective lens development associated with ocu-lar anomalies in C57BL/6 mice. Exp Eye Res 56:7–16.

Rupp RA, Snider L, Weintraub H. 1994. Xenopus embryos regulate thenuclear localization of XMyoD. Genes Dev 8:1311–1323.

Sedman SA, Mertz JE. 1988. Mechanisms of synthesis of virion proteinsfrom the functionally bigenic late mRNAs of simian virus 40.J Virol 62:954–961.

Sidow A, Bulotsky MS, Kerrebrock AW, Bronson RT, Daly MJ, ReeveMP, Hawkins TL, Birren BW, Jaenisch R, Lander ES. 1997. Serrate2is disrupted in the mouse limb-development mutant syndactylism.Nature 389:722–725.

Silver J. 1977. Abnormal development of the suprachiasmatic nuclei ofthe hypothalamus in a strain of genetically anophthalmic mice.J Comp Neurol 176:589–606.

Silver J, Hughes AF. 1974. The relationship between morphogeneticcell death and the development of congenital anophthalmia.J Comp Neurol 157:281–301.

Silver J, Puck SM, Albert DM. 1984. Development and aging of the eyein mice with inherited optic nerve aplasia: Histopathological stud-ies. Exp Eye Res 38:257–266.

Smith RS, Roderick TH, Sundberg JP. 1994. Microphthalmia and asso-ciated abnormalities in inbred black mice. Lab Anim Sci 44:551–560.

Snell GD. 1941. Inbred strains. Mouse Genetic News 1:9.Taylor BA, Navin A, Phillips SJ. 1994. PCR-amplification of simple

sequence repeat variants from pooled DNA samples for rapidlymapping new mutations of the mouse. Genomics 21:626–632.

Tsujino S, Rubin LA, Shanske S, DiMauro S. 1994. An A-to-C substitutioninvolving the translation initiation codon in a patient with myo-phosphorylase deficiency (McArdle’s disease). Hum Mutat 4:73–75.

Turner DL, Weintraub H. 1994. Expression of achaete-scute homolog 3in Xenopus embryos converts ectodermal cells to a neural fate.Genes Dev 8:1434–1447.

52 TUCKER ET AL.

Voronina VA, Mathers PH. 2000. Mutations in the human RX gene in apatient with anophthalmia. Invest Ophthalmol Vis Sci 41:S32.

Waye JS, Eng B, Patterson M, Barr RD, Chui DH. 1997. De novomutation of the beta-globin gene initiation codon (ATG3AAG) ina Northern European boy. Am J Hematol 56:179–182.

Webster EH Jr., Silver AF, Gonsalves NI. 1984. The extracellular matrixbetween the optic vesicle and presumptive lens during lens morpho-genesis in an anophthalmic strain of mice. Dev Biol 103:142–150.

Wilkens H. 1971. Genetic interpretation of regressive evolutionaryprocesses: Studies on hybrid eyes of two Astyanax cave popula-tions (Characidae, Pisces). Evolution 25:530–544.

Wright S. 1968. Evolution and the genetics of populations. Vol. I.Genetic and biometric foundations. Chicago: The University ofChicago Press.

Yamamoto Y, Jeffery WR. 2000. Central role for the lens in cave fisheye degeneration. Science 289:631–633.

Zhang L, Mathers PH, Jamrich M. 2000. Function of Rx, but not Pax6,is essential for the formation of retinal progenitor cells in mice.Genesis 28:135–142.

Zwaan J, Silver J. 1983. Crystallin synthesis in the lens rudiment of astrain of mice with congenital anophthalmia. Exp Eye Res 36:551–557.

53RX/RAX MUTATION IN THE EYELESS (EY1) MOUSE